Unified Static–Dynamic Pruning for Efficient LLM Inference

Summary: SPDP unifies static and input-adaptive dynamic pruning for LLM GPUs via Tiled-CBC storage and phase-specific CUDA/Tensor-Core kernels. It sustains efficient sparse execution, achieving up to 2.51× speedup and 25% higher sparsity than prior frameworks. (summarized by gpt-5.6-luna on Aug 28 2026)

Paper ID: hdd1a46ed64d70adb
Venue: VLDB
Year: 2026
Pagerank: 4.9793485e-05
Overall Rank: 10,838 | 27.14%
DOI: 10.14778/3836663.3836665

Incoming Non-self Citations Over Time

No non-self incoming citations found for this paper in this database.

Authors

1. Jinhyeok Kim (Seoul National University)
2. Yejoon Lee (Seoul National University)
3. Jaeyoung Do (Seoul National University)

BibTeX Citation

@article{kim_vldb26,
        title = {{Unified Static–Dynamic Pruning for Efficient LLM Inference}},
        author = {Kim, Jinhyeok and Lee, Yejoon and Do, Jaeyoung},
        journal = {PVLDB},
        series = {{VLDB} '26},
        volume = {19},
        number = {11},
        pages = {2950--2963},
        doi = {10.14778/3836663.3836665},
        url = {https://doi.org/10.14778/3836663.3836665},
        year = {2026}
}

Incoming Citations (Sorted by Pagerank)

Showing 0 of 0 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank

Outgoing Citations (Sorted by Pagerank)

Showing 1 of 1 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
4,351	Flash-LLM: Enabling Cost-Effective and Highly-Efficient Large Generative Model Inference with Unstructured Sparsity	2024	VLDB	6.642803e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	6,860	SpareLLM: Automatically Selecting Task-Specific Minimum-Cost Large Language Models under Equivalence Constraint	2025	SIGMOD
2	8,888	Optimized Batch Prompting for Cost-effective LLMs	2025	VLDB
3	5,161	HotPrefix: Hotness-Aware KV Cache Scheduling for Efficient Prefix Sharing in LLM Inference Systems	2026	SIGMOD
4	10,776	Efficient Cooperation-Aware Key and Value Management for LLM Inference	2026	VLDB
5	6,166	Apt-Serve: Adaptive Request Scheduling on Hybrid Cache for Scalable LLM Inference Serving	2025	SIGMOD
6	10,472	KVDrive: A Holistic Multi-Tier KV Cache Management System for Long-Context LLM Inference	2026	SIGMOD
7	6,560	RetroInfer: A Vector Storage Engine for Scalable Long-Context LLM Inference	2026	VLDB
8	9,158	Database Perspective on LLM Inference Systems	2025	VLDB
9	10,407	AlignedServe: Orchestrating Prefix-aware Batching to Build a High-throughput and Computing-efficient LLM Serving System	2026	SIGMOD
10	4,351	Flash-LLM: Enabling Cost-Effective and Highly-Efficient Large Generative Model Inference with Unstructured Sparsity	2024	VLDB